|
GOVERNANCE
Governance
Governance is not a slide deck. It is decision rights, audit trails,
and being able to answer “who approved this?” after something goes wrong.
I treat it as product work: the confirmation button, the escalation path,
the review queue, the log nobody wants until they need it. Policy PDFs matter;
the surface people click matters more.
|
|
| Questions I put on the wall |
- Who can override a model suggestion — and is that override logged?
- Where does harm bite hardest: money moves, health tips, public statements?
- Who decides when the model is unsure? Silence is not consent.
- Can a retrospective name the person, the rule, and the timestamp — without folklore?
- Is oversight in the interface, or only in a PDF nobody opens?
|
|
|
TWO ROOMS
“Excitement answers ‘can it?’ Governance answers ‘what happens when it fails?’”
Teams that only celebrate demos get responsibility fog.
Teams that only write policy get products nobody can operate.
The craft is joining those rooms before launch — especially in fintech and healthtech.
|
|
| Doorplates |
|
Clear doorplates make people more willing to walk in.
Ambiguity is what makes AI feel unsafe in practice —
not the existence of the model itself.
Home base for the writing:
Governed by Design.
|
|
|
|
|
| Special Offer |
|
FREE
Intro call / coffee chat.
No hard sell. Just talk shop
— agents, governance, product.
|
|
|
|
|
|
|
|
|
|
|
|
|
N
Optimized for
Netscape Navigator
|
|
|
Mac
Platinum UI
c. 1996
|
|